Method for detecting living beings in a vehicle interior and vehicle

Abstract

Method for detecting living beings (1) in a vehicle interior (2), wherein the vehicle interior (2) is acoustically monitored by at least two microphones (3) and a computing unit (4) evaluates signals generated by the microphones (3) using artificial intelligence, characterized by the following method steps: a) Implementing a NAF database (5) in the computing unit (4), wherein the NAF database (5) comprises a plurality of Neural Acoustic Fields (NAF) of a vehicle interior for different presence scenarios, and wherein the NAF database (5) is generated by an initial measurement campaign in the representative vehicle interior by recording at least one Neural Acoustic Field (NAF) each time while varying at least dummy type, dummy number and dummy seating position; b) Recording a soundscape in the vehicle interior (2) using the microphones (3) at least when vehicle doors are closed; c) Comparison of the recorded soundscape with the Neural Acoustic Fields (NAF) stored in the NAF database (5) by the computing unit (4); d) Determining at least one Neural Acoustic Field (NAF) which shows the highest correspondence with the recorded soundscape by the processing unit (4); and e) Determining the presence of living beings (1) in the vehicle interior (2) by the computing unit (4), corresponding to the respective presence scenario underlying the Neural Acoustic Field (NAF) determined in step d).

Description

[0001] The invention relates to a method for detecting living beings in a vehicle interior according to the type defined in more detail in the preamble of claim 1 and to a vehicle according to the type defined in more detail in the preamble of claim 10.

[0002] In everyday life, it can happen that children or animals are forgotten or deliberately left behind in the vehicle interior. For example, a dog owner wants to do a quick shopping trip but doesn't want to take their dog with them. Particularly in strong sunlight in summer and cold temperatures in winter, this can lead to life-threatening situations, as the vehicle interior can heat up or cool down significantly, putting a child or pet at risk of overheating or hypothermia. This creates a need for means and methods to warn vehicle users about children or animals left behind in the vehicle.

[0003] For example, DE 10 2019 212 412 A1 discloses a monitoring device and a method for monitoring a living being in the interior of a vehicle. The document describes the monitoring of the vehicle interior using visual detection devices such as cameras, acoustic detection devices such as microphones, and temperature sensors. Data generated by said cameras, microphones, and temperature sensors is evaluated by a computing unit using artificial intelligence. The computing unit performs a classification to determine the respective presence situation. To determine whether living beings are present in the vehicle interior, data from additional sensors such as seat occupancy sensors, motion sensors, and / or seat belt sensors can also be considered. A baby monitor function can be provided through the use of cameras and microphones, as well as through the presence of display devices and loudspeakers in the vehicle.This allows video and audio content recorded in the vehicle interior to be transmitted to the vehicle user's smartphone, and audio and video content recorded from the smartphone can be transmitted back to the vehicle. This allows the vehicle user to reassure children present in the vehicle interior. However, due to the requirement to install various sensor types, the corresponding monitoring device is comparatively complex and expensive.

[0004] It would therefore be desirable to provide a function for detecting the presence of living beings in vehicle interiors using structurally simple and thus cost-effective means.

[0005] The detection of distance information using sound is well known. A wide variety of devices exist for this purpose, such as echo sounders, sonar, or ultrasonic sensor systems such as those used in parking aids, colloquially known as "parking beepers."

[0006] The spatial modeling of the acoustic characteristics of an environment is also the subject of current research. Andrew Luo et al. describe methods and means by which it is possible to obtain structural information about a scene purely using acoustics. To this end, the researchers introduce the concept of neural acoustic fields (NAFs), referred to in their work as neural acoustic fields. Such a neural acoustic field contains geometric information that describes the structure of the measured environment. To generate such a neural acoustic field, sound measurements are taken in an environment to be mapped, for example, a room or an apartment. These sound measurements require the performance of a large number of individual measurements.For each individual measurement, a sound source and a sound receiver are randomly placed in the environment under investigation. A sound pulse is emitted by the sound source, which is then detected by the sound receiver. The measured impulse response is converted into a spectrogram using a short-time Fourier transform. The resulting spectrograms are then processed by an artificial neural network, which determines the desired neural acoustic field as the output variable. For details, see: Learning Neural Acoustic Fields, Andrew Luo, Yilun Du, Michael J. Tarr, Joshua B. Tenenbaum, Antonio Torralba, Chuang Gan, https: / / doi.org / 10.48550 / arXiv.2204.00628.

[0007] The use of a machine learning model to recognize the 3D geometry of a room from sound recordings is disclosed in the publication VAN DEN HEUVEL, M.: Open your ears: A machine learning system simulates how humans hear. November 3, 2022. URL: https: / / www.ingenieur.de / technik / fachbereiche / ittk / ohren-auf-ein-maschinelles-lernsystem-simuliert-wie-menschen-hoeren.

[0008] DE 102 46 048 A1 describes a seat occupancy detection system for a vehicle based on an acoustic transfer function determined using the speakers of a vehicle's hi-fi system and at least one microphone. Seat occupancy is determined by regularly measuring the acoustic properties embodied in the acoustic transfer function and comparing them with stored reference transfer functions.

[0009] The present invention is based on the object of providing an improved method for detecting living beings in a vehicle interior, which also functions reliably with simple hardware.

[0010] According to the invention, this object is achieved by a method for detecting living beings in a vehicle interior having the features of claim 1. Advantageous embodiments and further developments as well as a vehicle for carrying out the method emerge from the dependent claims.

[0011] A generic method for detecting living beings in a vehicle interior, wherein the vehicle interior is acoustically monitored by at least two microphones and a computing unit evaluates signals generated by the microphones using artificial intelligence, is further developed according to the invention by the following method steps: a) Implementing a NAF database in the computing unit, wherein the NAF database comprises a plurality of neural acoustic fields of the vehicle interior for different presence scenarios, and wherein the NAF database is generated by an initial measurement campaign in a representative vehicle interior by recording at least one neural acoustic field while varying at least the dummy type, dummy number and dummy seating position; b) recording the noise in the vehicle interior using the microphones at least when the vehicle doors are closed; c) Comparison of the recorded background noise with the neural acoustic fields stored in the NAF database by the computing unit; d) Determining at least one neural acoustic field which has the highest correspondence with the recorded background noise by the computing unit; and e) Determining the presence of living beings in the vehicle interior by the computing unit, corresponding to the respective presence scenario underlying the neuronal acoustic field determined in step d).

[0012] The method according to the invention enables acoustic monitoring of the vehicle interior in a completely novel way, allowing the presence of living beings in the vehicle interior to be reliably and easily detected. The neural acoustic fields represent an acoustic mapping of the vehicle interior. To detect living beings in the vehicle interior, corresponding maps are generated for all possible presence scenarios. These measurement campaigns are carried out, for example, by the vehicle manufacturer.

[0013] Dummies such as crash dummies, dolls, water-filled canisters, and the like are used to simulate the presence of living beings. Adults, children, and animals can be represented using different dummy types, such as larger or smaller dummies. A dummy can also be covered with a carpet, feathers, or hair to simulate fur or feathers. The number of dummies varies between zero and the maximum number of seats in the vehicle interior. The dummies are arranged in all possible seating positions. This ensures that all possible presence scenarios are covered by the neural acoustic fields.

[0014] If the vehicle is later in use and there is, for example, an adult driver in the driver's seat and a child and a dog in the rear of the vehicle, this results in corresponding acoustics in the vehicle interior. This acoustics is mapped by a corresponding neural acoustic field. In this case, there is a correlation to a neural acoustic field from the initial measurement campaign with the following presence scenario: a large dummy in the driver's seat and a small and an extremely small dummy in the rear of the vehicle on the left and right seats. In the NAF database, information is linked to the respective neural acoustic field, indicating the type of dummy, the number of dummies, and the dummy seating position underlying the respective presence scenario.

[0015] To ensure the highest possible agreement between the measured noise level and a neural acoustic field stored in the NAF database, the corresponding measurement campaigns are conducted individually for various representative vehicles. "Representative" in this case means that the vehicle used for the initial measurement campaign corresponds to at least the same model series as the vehicle in which the method is ultimately applied. This ensures that the geometry of the vehicle interior is essentially the same, which is important for the acoustic properties of the vehicle interior.

[0016] The initial measurement campaign can be designed as comprehensively as desired. This allows for the investigation of other influencing factors such as seat adjustment, in particular the seat backrest inclination, the position or movement of an entire vehicle seat, the presence of child seats, and / or other objects carried in the vehicle. In addition to the dummy type, number of dummies, and dummy seat position, corresponding neural acoustic fields can be recorded for the various seat adjustments, as well as for, for example, a handbag placed in the rear of the vehicle, a laptop bag lying on the passenger seat, and the like.

[0017] For each parameter combination, at least one neural acoustic field is recorded during the initial measurement campaign. Preferably, however, multiple neural acoustic fields can also be recorded, which can reduce measurement inaccuracies or errors resulting from statistical events. In particular, for the same presence scenario, multiple neural acoustic fields are recorded for different sound source configurations. For this purpose, the type of sound source and / or the sound source position relative to the vehicle interior are varied.

[0018] It is particularly practical if the closing of the vehicle doors is used as a sound source later on to record the background noise in the vehicle interior. The closing of the vehicle doors is a particularly suitable sound source because it generates a short and concise sound pulse. Furthermore, it is not necessary to emit an artificial excitation signal in the vehicle interior. "When closing" in this context means that the sound signals relevant for processing are analyzed within a period of time from the moment the vehicle doors are closed until shortly afterwards. This period can be on the order of a few seconds or even milliseconds. The microphones must be activated at least for the duration of the recording. The microphones can also be activated before the actual recording period and deactivated some time after the recording period has ended.For example, the microphones can be activated when sensors detect the opening of the vehicle doors. When the vehicle doors open and close, the vehicle is usually stationary. This is particularly the moment when vehicle occupants enter or exit the vehicle. This is a relevant moment for determining the presence of living beings in the vehicle interior. However, it is also generally possible that the microphones are activated again while driving, and audio signals are recorded. This will be discussed further below.

[0019] By installing multiple microphones in the vehicle interior, multiple neural acoustic fields can be recorded simultaneously from a single acoustic signal source. The installation position of the corresponding microphones between the representative vehicle and the vehicle in which the method is being performed is accordingly consistent. Vehicles typically have several such interior microphones, for example, to provide a hands-free system for telephony or to capture voice commands for a voice assistant. To optimally capture the voices of different vehicle occupants, microphones are typically installed at different locations.

[0020] In step d), the computing unit determines at least one neural acoustic field that best matches the recorded background noise. The comparison of the background noise described in step c) can, in particular, provide for a neural acoustic field to be generated from the recorded background noise and compared with the NAF database. If such a field stands out from the entire set of neural acoustic fields stored in the NAF database, it becomes clear which presence scenario is present, i.e., how many living beings are in the vehicle interior, where they are sitting, and what type of living being it is, for example, an adult, a child, or an animal.

[0021] By processing the acoustic signals, the computing unit can also determine which animal is involved, for example by recognizing a bark, meow, chirp or the like.

[0022] However, cases can also occur in which several neural acoustic fields from the NAF database show similarity to the recorded background noise. The computing unit then compares the information associated with the respective presence scenarios. For example, the neural acoustic fields may show a high degree of similarity where two living beings are present in the vehicle interior. However, the respective seating position, for example, varies. In such a case, the method according to the invention can be used to determine the number of living beings present in the vehicle interior, but not the exact seating position.

[0023] However, with the aid of the method according to the invention it is at least possible to distinguish whether at least one living being is present in the vehicle interior or not.

[0024] An advantageous development of the method according to the invention provides that the slamming of a vehicle door serves as the acoustic signal source in the initial measurement campaign for generating the NAF database, wherein, in particular, neural acoustic fields are also recorded under variation of the slammed vehicle door. Since the method according to the invention for determining the presence of living beings in the vehicle interior is carried out at least when the vehicle is stationary, when the vehicle doors are opened and closed, and the slamming of the vehicle door(s) serves as the sound source(s), the slamming of the vehicle doors also serves as the sound source for generating the NAF database in a particularly advantageous manner in the initial measurement campaign. Thus, the neural acoustic fields generated in the initial measurement campaign are based on the same sound source that is also used later when carrying out the method in the vehicle.This makes it possible to increase the degree of correspondence between the background noise recorded in the vehicle interior (in particular a neural acoustic field derived from the background noise) and the neural acoustic fields stored in the NAF database.

[0025] According to a further advantageous embodiment of the method, the computing unit determines which vehicle door was closed when the vehicle door is closed, taking into account a door closing sensor. Depending on which vehicle door is closed and thus serves as the sound source, this influences the corresponding neural acoustic field due to the changed position of the sound source. Thus, neural acoustic fields are recorded individually for each possible combination of slammed vehicle doors. A single vehicle door can be slammed, or several vehicle doors can be slammed simultaneously or in quick succession. This is taken into account accordingly in the initial measurement campaign. Taking into account the data transmitted by a door closing sensor, the computing unit can precisely determine which vehicle door is slammed.The processing unit can then use this information to determine the correspondence between the background noise recorded in the vehicle interior and the neural acoustic fields stored in the NAF database. The processing unit selects only those neural acoustic fields from the NAF database that are suitable for a match in which the corresponding vehicle door served as the sound source.

[0026] A further advantageous embodiment of the method according to the invention further provides that, depending on the vehicle equipment, different initial measurement campaigns are carried out to generate the NAF database. In addition to the influencing factors already mentioned, the vehicle equipment also affects the respective neural acoustic fields. For example, if a certain type of vehicle seat is installed, such as a sports seat, or if it has a certain material composition such as leather, or if other special components are arranged in the vehicle interior, such as burl wood trim, this can influence the soundscape in the vehicle interior. For example, sound is reflected differently by a smooth surface than by a rough surface such as velvet or textured plastic.

[0027] The vehicle manufacturer can therefore conduct corresponding measurement campaigns not only for different model series, but also for different vehicle equipment within each model series. All neural acoustic fields generated in this way can then be implemented in the vehicle's processing unit, or only those recorded in a representative vehicle interior with the corresponding vehicle equipment. This saves storage space in the processing unit. If the vehicle equipment changes over the vehicle's lifetime, however, any missing NAF database or neural acoustic fields must be subsequently incorporated into the processing unit. This can be done, for example, via an update, preferably a wireless "over-the-air" update.

[0028] According to a further advantageous embodiment of the method according to the invention, the computing unit carries out a measurement campaign to expand the NAF database during the vehicle's operational phase, whereby information describing a current presence scenario is proactively requested from a user. In other words, the computing unit is thus capable of learning. The vehicle user can record new neural acoustic fields during the operational phase, i.e., related to the vehicle's life cycle, the period in which the vehicle is used by a user. For this purpose, dummies are no longer used, but rather the living beings that typically travel with the vehicle, such as family members or pets. This further increases the degree of correspondence between the noise levels recorded in the vehicle and the neural acoustic fields stored in the NAF database.In order for the computing unit to be able to communicate the corresponding presence scenario, this information must be manually entered into the computing unit. The user can use a suitable human-machine interface, such as a touch-sensitive display, for this purpose.

[0029] A further advantageous embodiment of the method according to the invention further provides that method steps b) to e) are carried out again at a later time after the vehicle door has been closed, wherein an acoustic excitation signal is output via at least one loudspeaker arranged in the vehicle. As already mentioned, the presence of living beings in the vehicle interior can be detected using the method according to the invention not only when the doors are slammed, but also at a later time, for example while the vehicle is moving. This makes it possible to verify the previously determined presence of living beings, in particular the number, type of living being and seat positioning. Since vehicle doors should not be opened and closed while the vehicle is moving, a different sound source is preferably used.The vehicle's loudspeakers are suitable for this purpose, particularly audio signals are output via loudspeakers arranged at various locations in the vehicle interior. Here, too, the excitation signals preferably match those used in the initial measurement campaign.

[0030] For example, a pet can be passed into the vehicle through an open door window while the vehicle is stationary. This allows the correct count of the pet to be determined while driving.

[0031] Preferably, the acoustic excitation signal emitted via the at least one loudspeaker exclusively comprises frequencies in the ultrasonic frequency band and / or the infrasonic frequency band. This has the advantage that corresponding excitation signals are not perceptible to the vehicle occupants. This enables the method according to the invention to be carried out without the vehicle occupants noticing. This particularly improves user comfort.

[0032] A further advantageous embodiment of the method according to the invention further provides that the computing unit in method step d) also uses artificial intelligence to determine the neural acoustic field that corresponds to the background noise. Proven methods and procedures from artificial intelligence, such as artificial neural networks, in particular deep artificial neural networks, can be used for this purpose. This allows for even more reliable determination of correspondences between the recorded background noise and the neural acoustic fields stored in the NAF database. In particular, a corresponding machine learning model compares a neural acoustic field derived from the recorded background noise with the NAF database.

[0033] According to a further advantageous embodiment of the method according to the invention, the computing unit only initiates an emergency measure once a living being has been detected in the vehicle interior and the vehicle subsequently receives a door locking signal. Various actions can be performed as emergency measures, such as activating lighting devices, for example, the headlights, taillights, a direction indicator, and the like, as well as issuing acoustic warning signals, for example, via the vehicle's horn. This allows pedestrians in the vicinity of the vehicle to be warned. Warning messages can also be transmitted via a wireless communication connection, for example, to a mobile device of the vehicle owner or to an emergency service.Furthermore, the computing unit can control a vehicle function, for example, opening a window by operating the power windows or activating the vehicle's air conditioning system. This can prevent overheating in the summer, for example.

[0034] By only activating the emergency measure once the vehicle is locked, this prevents unnecessary emergency measures from being initiated. If the vehicle doors are unlocked, this typically means that a driver is present or nearby. Therefore, children and animals inside the vehicle are generally not at risk. However, if the vehicle is locked, the driver will also leave, increasing the risk to children and / or animals left behind in the vehicle. This is taken into account.

[0035] In a vehicle comprising at least two microphones and a computing unit, the microphones and the computing unit are configured according to the invention to carry out a method described above. The vehicle can be any vehicle, such as a car, truck, van, bus, or the like. The microphones are arranged at various positions in the vehicle interior. This allows for particularly comprehensive and differentiated recording of various neural acoustic fields. The computing unit can be a central on-board computer, a control unit of a vehicle subsystem, a telematics unit, or the like. Various steps to be carried out by the computing unit can also be distributed across multiple physically separate computing units.The one or more computing units comprise computer-readable storage media on each of which a computer program product is stored, the execution of which by a processor causes the method steps explained in connection with the method according to the invention to be implemented in the vehicle.

[0036] Further advantageous embodiments of the method according to the invention for detecting living beings in a vehicle interior and of the vehicle also emerge from the exemplary embodiments which are described in more detail below with reference to the figures.

[0037] Showing: Fig. 1 a schematic representation of a neural acoustic field of an apartment; and Fig. 2 a schematic representation of the sequence of a method according to the invention for determining the presence of living beings in a vehicle interior based on the generation of neuronal acoustic fields.

[0038] A neural acoustic field (NAF) corresponds to a representation of the acoustic properties of a measured environment. Andrew Luo et al. describe how such neural acoustic fields (NAFs) can be generated in their paper "Learning Neural Acoustic Fields" (https: / / doi.org / 10.48550 / arXiv.2204.00628). Using such a neural acoustic field (NAF), the structures, i.e., the geometry of the space under investigation, can be estimated. Thus, geometric information can be derived purely based on acoustic measurements.

[0039] Fig. Figure 1 shows such a neural acoustic field (NAF) of an apartment 6. It depicts a sound source 7 placed at an arbitrary location within the environment to be examined, in this case, the apartment 6. With the help of acoustic recording devices, for example a (possibly stereo) microphone, an acoustic signal emitted by the sound source 7 is recorded. Sound waves 8 propagating in the environment to be measured are also shown. Solid lines represent sound emanating from the sound source 7, and dashed lines represent echoes reflected back from the environment. The environment to be examined is mapped by varying the location of the sound source 7 and the recording devices (not shown in detail). Impulse responses recorded by the acoustic recording devices are subjected to a short-time Fourier transform, and a spectrogram is generated from this.The spectrograms of a large number of individual measurements are then provided as input to a machine learning model in the form of an artificial neural network. The artificial neural network then determines the . Fig. 1 depicted neural acoustic field (NAF). Such a neural acoustic field (NAF) shows the loudness or sound pressure level as a function of the position of the sound source 7 used as a spatial distribution of the examined environment. Fig. 1, loudness is represented as a grayscale, with light gray corresponding to loud sounds and dark gray or black to quiet sounds. Gaps 9, which correspond to walls or objects such as furniture, are clearly visible.

[0040] This technique, described in the paper by Andrew Lou et al., can be used to detect living beings in vehicle interiors. This is possible using a method according to the invention, the process of which is described in Fig. 2 is shown schematically.

[0041] First, a vehicle interior 2 must be measured in a measurement campaign. For this purpose, dummies of various designs are placed in different seating positions in the vehicle interior 2, and various sound excitation signals are output to measure the background noise in the vehicle interior 2. A multitude of neural acoustic fields (NAF) are then generated for different presence scenarios and aggregated in an NAF database 5. This NAF database 5 is then incorporated into a computing unit 4 of the vehicle, which is intended to implement the method according to the invention.

[0042] During later use of the vehicle, the background noise when the vehicle doors are closed is recorded by several microphones 3, for example four microphones 3 distributed throughout the vehicle interior 2, in order to determine the presence of living beings in the vehicle interior 2. In the Fig. In the embodiment shown in Figure 2, the living beings 1 in the vehicle interior 2 are an adult driver in the driver's seat and a child in a right rear seat. The presence of said living beings 1 influences the propagation of sound waves 8 in the vehicle interior 2 when the vehicle door is closed. The vehicle door thus serves as the sound source 7.

[0043] The processing unit 4 can derive an individual neural acoustic field (NAF) or an averaged neural acoustic field (NAF) for each microphone 3 from the thus recorded background noise. This is shown in Fig. 2 is shown at the top right and shows corresponding gaps 9 at the locations where structures are located in the vehicle interior 2. Shown is a section of the neural acoustic field (NAF) at a specific geodetic height. This geodetic height corresponds in particular to the height at which the respective microphones 3 are mounted in the vehicle interior 2. It is particularly advantageous to arrange several microphones 3 at different geodetic heights in the vehicle interior 2 so that sections can be generated at different height positions. In addition to the living beings 1, gaps 9 are shown for the corresponding backrests of the vehicle seats.

[0044] This in Fig.The process shown in Figure 2 can be implemented explicitly or implicitly. An explicit process requires that the neural acoustic field (NAF) derived from the background noise is geometrically compared with the neural acoustic fields (NAF) stored in the NAF database 5. For example, a correlation of overlapping lines is mathematically determined. An implicit process, on the other hand, requires that the respective neural acoustic fields (NAF) are merely examined for "similarity." For this purpose, artificial intelligence, in particular, can compare the respective data structures of the neural acoustic fields (NAF). Such a comparison using a corresponding machine learning model then runs as a black box.

[0045] To compare the background noise or the neural acoustic fields (NAF) derived from the background noise with the NAF database 5, the computing unit 4 checks which neural acoustic field (NAF) stored in the NAF database 5 has the highest match. To do this, the computing unit 4 reads the corresponding NAF database 5. Different match probabilities are shown as examples. At least the neural acoustic field (NAF) with the highest match is selected as the match in a result 10.

[0046] Accordingly, the computing unit 4 deduces that the presence scenario of the respective neural acoustic field (NAF) present during the initial measurement campaign also currently applies to the vehicle. This makes it possible to determine at least whether a living being 1 is present in the vehicle interior 2 or not. The more clearly a specific neural acoustic field (NAF) from the NAF database 5 matches the neural acoustic field (NAF) derived from the background noise, the more information can be derived, in particular the number of living beings 1, the type of living being 1 (i.e., adult, child, animal, or similar), and the seat occupied by living being 1.

Claims

[1] Method for detecting living beings (1) in a vehicle interior (2), wherein the vehicle interior (2) is acoustically monitored by at least two microphones (3) and a computing unit (4) evaluates signals generated by the microphones (3) using artificial intelligence, characterized by the following procedural steps: a) Implementing a NAF database (5) in the computing unit (4), wherein the NAF database (5) comprises a plurality of neural acoustic fields (NAF) of a vehicle interior for different presence scenarios, and wherein the NAF database (5) is generated by an initial measurement campaign in the representative vehicle interior by recording at least one neural acoustic field (NAF) in each case while varying at least the dummy type, dummy number and dummy seating position; b) recording a background noise in the vehicle interior (2) by means of the microphones (3) at least when closing vehicle doors; c) comparing the recorded background noise with the neural acoustic fields (NAF) stored in the NAF database (5) by the computing unit (4); d) determining at least one neural acoustic field (NAF) which has the highest correspondence with the recorded background noise by the computing unit (4); and e) Determining a presence of living beings (1) in the vehicle interior (2) by the computing unit (4), corresponding to the respective presence scenario underlying the neural acoustic field (NAF) determined in step d). [2] Method according to claim 1, characterized bythat the slamming of a vehicle door serves as the acoustic signal source in the initial measurement campaign for generating the NAF database (5), wherein in particular neural acoustic fields (NAF) are recorded under variation of the slammed vehicle door. [3] Method according to claim 1 or 2, characterized by that the computing unit (4) determines which vehicle door was closed when the vehicle door is closed, taking into account a door closing sensor. [4] Method according to one of claims 1 to 3, characterized by that, depending on the vehicle equipment, different initial measurement campaigns are carried out to generate the NAF database (5). [5] Method according to one of claims 1 to 4, characterized bythat the computing unit (4) carries out a measurement campaign to expand the NAF database (5) in a deployment phase of the vehicle, wherein information describing a current presence scenario is proactively requested from a user. [6] Method according to one of claims 1 to 5, characterized by that the method steps b) to e) are carried out again at a later time after the vehicle door has been closed, an acoustic excitation signal being emitted via at least one loudspeaker arranged in the vehicle. [7] Method according to claim 6, characterized by that the acoustic excitation signal emitted via the at least one loudspeaker comprises exclusively frequencies in the ultrasonic frequency band and / or infrasonic frequency band. [8] Method according to one of claims 1 to 7, characterized bythat the computing unit (4) in method step d) also uses artificial intelligence to determine the neural acoustic field (NAF) corresponding to the background noise. [9] Method according to one of claims 1 to 8, characterized by that the computing unit (4) only carries out an emergency measure when a living being (1) has been detected in the vehicle interior (2) and the vehicle then receives a door locking signal. [10] Vehicle comprising at least two microphones (3) and a computing unit (4), characterized by that the microphones (3) and the computing unit (4) are arranged to carry out a method according to one of claims 1 to 9.

Images